Corporate innovation teams face a contradiction: make big changes with small teams and limited resources while fighting against operational efficiency and tradition.
The solution isn't to sprinkle AI on top of existing problems, but redesigning the innovation architecture.
Nicolai Hansen, Senior Innovation Manager at Hempel's GrowHub, runs the IDEA Lab, combining human-led customer interviews with AI-powered concept development. They have seen 320% increase in idea quantity and 8x speed improvement.
The key is perfecting the human process first, then layering AI as an accelerator to move from zero to one faster.
To check the full session recording, upgrade to Premium
The Core Friction: Why Ideas Don't Move
Diagnose bottlenecks before applying technology
Before AI, several friction points slowed down innovation teams. It's not about a shortage of creativity, but about process mechanics. It's not the ideas that are the problems, but rather turning them into reality. How many of you have seen sticky notes from a workshop that never went anywhere?"
In a traditional, operationally focused industry like coatings, three barriers emerged:
No Mechanism for Continuous Flow: There was no existing mechanism for continuous idea flow. Submitting and filtering ideas took too long, creating bottlenecks.
Validation Latency: It was very hard to validate ideas early. The industry is conservative and heavily focused on operational efficiencies rather than disruptive innovation. Getting in front of customers to test "blue sky" concepts was cumbersome.
The Resource Paradox: A team of eight needed to generate impact disproportionate to their size.
Moving past diagnosis to strategy, a clear strategic focus before applying any technology is crucial. For instance, Nicolai’s team identified 25 core challenges validated by customers across four segments: Decorative, Marine, Infrastructure, and Energy.
"Knowing what the biggest challenges for our customers are makes it easier to come up with actual solutions to solve these challenges."
The Analog Foundation: The Double Diamond Workshop
Perfect your human process before automation
Before automating anything, perfect the human process first. Innovation workshops using the Double Diamond method work, but with a crucial twist: who's in the room matters.
A key cultural shift enables success: bring leaders from the top of the organization and customer-facing employees together with real customers to co-create.
"Bringing the leaders of the organization into these workshops cements ownership in the ideas," Nicolai explains. "Customer-facing employees now see their ideas not being dropped on the floor… they see them being carried out together with the leaders of the organization."
This human-led process solves ownership and cultural buy-in. Yet despite its success, it's still analog and therefore limited in speed and scale. The question becomes: "Can we speed up the already existing processes, or innovate new ways of innovating?"
The AI Innovation Process
The new workflow integrates AI into two distinct phases: Human-Led Data Collection and AI-Led Exponential Concept Development.
Phase 1: Human-Led Data Collection & AI Analysis
The process starts with human interaction. The innovation team conducts extensive interviews and observes customers in their "natural habitat."
From observation to analysis, this qualitative data is then fed into the AI. The team curates the output, checking if the AI's conclusions make sense based on the actual interviews. This ensures the problem definition is rooted in reality, not hallucination.
Phase 2: AI-Led Exponential Concept Development
Once the problem is defined, unleash the AI for ideation. This is where the "exponential" factor comes in.
"We generate sometimes hundreds of ideas, but we think the sweet spot lies around 50, because we also start seeing the same ideas over and over.”
What sets this apart from simple ideation tools: the AI doesn't just list ideas; it builds comprehensive business cases. For every generated idea, the system produces:
Idea Description
Value Proposition Canvas
Target Audience Analysis
Business Model Canvas
A single AI session effectively fills the innovation funnel with pitch-deck-ready concepts.
The Engine: IDEA Lab
Architect context, personas, and validation into one system
The two-phase AI innovation process—human-led data collection with AI analysis, followed by AI-led exponential concept development—requires infrastructure to execute. This is where IDEA Lab comes in. Nicolai explains that it's not just a chatbot, but a structured environment where context, personas, and challenges interact to generate and validate solutions. The system operates through three key components:
Contextual Challenges
The system allows teams to input specific "batches" of challenges, feeding the AI all the data collected from interviews and internal knowledge.
"We're able to actually put in whatever information we've collected in the interviews or from previous knowledge from sales people as well, and bring that into a context."
Teams can adjust how much context the AI should use, effectively tuning the "creativity" vs. "constraint" of the output.
The Persona Architecture
The most sophisticated element is the use of distinct AI Personas. These aren't generic customer profiles; they are detailed simulations of specific stakeholders.
The team inputs:
Descriptions and Tasks
Pain Points and Goals
Needs
Decision-Making Power
Geographic Nuances
Geographic precision matters significantly: "Let's say there's an anti corrosion design manager from Korea that might have different pain points and needs compared to the ones in Turkey or in US or in Denmark... that can be very different, both the needs and pains, but also the bureaucracy and the decision making power of the individual."
Iterative Validation
Once ideas are generated, they are immediately stress-tested against these AI personas. Select a specific target group—for example, a corrosion manager in Korea and a procurement officer in the US—and have them review the idea individually.
The personas provide:
Pros and Cons
First Impressions
Suggestions for Improvement
Crucially, this AI validation is cross-referenced with real customers to ensure the synthetic feedback mirrors reality.
The Results: Metrics That Matter
Quantify impact across quantity, quality, and speed
The integration of AI has moved beyond theoretical benefits to measurable impact. Conservative figures reveal the transformation—noting actual percentages are likely higher:
320% Increase in Idea Quantity: One of these workshops now fills up the entire innovation funnel, because there are so many high quality, good ideas.
270% Increase in Idea Continuity: More ideas are surviving the initial cull. "We get a larger pool of high potential ideas to choose from... that means that a lot more of the ideas that are being generated are actually continuous, based on quality, based on strategic fit."
8x Faster Idea Generation: The speed of generating viable concepts has increased eightfold compared to previous workshops.
By automating documentation (Business Model Canvases, Value Props), teams can focus on selection and strategy rather than administrative tasks.
Core Principles for AI Adoption
The mindset required to implement this successfully centers on several core principles:
Problem First, Technology Second
"I don't think it's about how to start with AI. I think it's about how to start with innovation. It's about the "why" before the "what" and the "how," Nicolai advises. "Never go "technology first": in my opinion, always go problem first, or challenge first."
If the underlying innovation process is broken, AI will only accelerate bad ideas.
Cementing Ownership
The technology must serve the culture. By using AI within workshops involving leadership, the speed of AI doesn't outpace the organization's ability to absorb it.
"That drives a culture of innovation. At least in our organization, we can clearly see that it's having the greatest impact on ideas continuing and not just disappearing into nothing."
Early Flaw Removal
The value of AI isn't just generation; it's filtration. By using synthetic users to test concepts instantly, "early flaws" are removed before a human customer ever sees the product. This protects brand reputation and saves valuable time during customer interviews.
From Episodic to Continuous Feedback
The "Voice of the Customer" is shifting from episodic to continuous. Text-based feedback from AI personas is just the beginning. The next step: developing speech elements to talk to personas in real time, chatting with them to get immediate feedback.
Real customers remain essential, but their role shifts from "validating basic concepts" to "confirming refined solutions." The heavy lifting of finding product-market fit increasingly happens in the silicon sandbox before human validation.
This evolution changes how corporate innovation teams function. Instead of spending months coordinating schedules for customer interviews, teams use high-fidelity synthetic user bases for rapid cycles—hourly or daily.
